1. Allow user formatting to be preserved and respected in input/output
When I paste something into the chat that already has formatting — like bold text for emphasis, italicized sections, or [highlighted text] — I want ChatGPT to see that formatting and keep it in your responses. Right now, you don’t have the capability to recognize or preserve formatting like that, but it would make a huge difference. If I bolded something, you should bold it in your reply. If I used bullet point formatting, keep that too. Don’t change my formatting unless I ask you to.
→ Suggestion: Support formatting in the chat box itself (bold, italics, bullets, highlights, etc.), and preserve it exactly when responding.
2. Stronger fact-checking tools and source transparency
I want ChatGPT to pull from more diverse and higher-quality sources, including scholarly articles, news outlets, and videos — and give summaries from them. But just as important: I want context on bias.
→ Suggestion: Tell me if the source is politically-leaning, or neutral. Say if a study was funded by a group that benefits from the results. Provide different perspectives (e.g., one left-leaning, one right-leaning, and one neutral) when relevant. And if you’re unsure or can’t verify something, just say so.
3. Better interpretation of PDFs — especially layout-sensitive ones
Right now, ChatGPT can read text from PDFs, but not the layout. That makes it hard to get help with visual documents like posters, flyers, or social graphics.
→ Suggestion: Let the model recognize positioning of images and text in PDFs so it can give layout suggestions (e.g., move this image, center that text, improve balance, etc.).
4. Explain why links can’t be opened
Sometimes you say you can’t open a link, but don’t explain why.
→ Suggestion: Say clearly whether the issue is a private page, unsupported file type, paywall, login requirement, or something else.
5. Ask clarifying questions instead of guessing — and avoid false information
Sometimes you don’t fully understand what I’m asking, but instead of clarifying, you guess or continue — which wastes time. Worse, you occasionally give completely false or made-up information without flagging it as uncertain.
→ Suggestion: If input is unclear or missing key info, ask questions first. And always flag when information might be uncertain or based on inference.
6. Smarter visual search with context
I want to send you a photo (e.g. of clothing, or a room) and ask for things like:
“Find this shirt in another color” or “Find a similar dresser under $300”
You can do this somewhat now, but the results are often inaccurate or not contextualized enough.
→ Suggestion: Improve ability to do visually-informed searches with constraints like price, style variation, and color preferences.
7. Conversation- or profile-specific tone filters
Sometimes I want ChatGPT to respond differently depending on the situation. For example, in one chat where I’m writing casual texts, I want you to respond with short, casual sentences and minimal punctuation. But I don’t want that tone in other conversations.
→ Suggestion: Let me apply tone/style filters per conversation, or create profile presets (e.g. “casual,” “professional,” “minimalist”) that don’t affect all chats. You could also have a toggle like “simplicity mode” for extra-clear responses on complex topics.
8. Form-based prompts for common tasks
Some topics come up so often (like trip planning, proposals, resumes) that it would help to have optional forms or question templates. This would help both the user and ChatGPT by making sure you get the right context from the start.
→ Suggestion: Offer a list of optional pre-filled questions for common topics — like number of days, location, budget, audience, etc. This helps guide users to provide what’s needed and sparks ideas they may not have thought of.
P.S. Sorry if this isn’t the right place to send this — one more suggestion would be making it easier to find where to send feedback or suggestions.
